2015/12/10 by Walid Shalaby, Shalaby, Walid, Wlodek Zadrozny +1
Computer Science · #Advanced Text Analysis Techniques #Computation and Language (cs.CL) #FOS: Computer and information sciences #H.3.1 #Text and Document Classification Technologies #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1512.03465
openalex publication_date 2015/12/10 · openalex created_date 2022/10/06 · openalex updated_date 2026/07/28
Mined Semantic Analysis (MSA) is a novel concept space model which employs\nunsupervised learning to generate semantic representations of text. MSA\nrepresents textual structures (terms, phrases, documents) as a Bag of Concepts\n(BoC) where concepts are derived from concept rich encyclopedic corpora.\nTraditional concept space models exploit only target corpus content to\nconstruct the concept space. MSA, alternatively, uncovers implicit relations\nbetween concepts by mining for their associations (e.g., mining Wikipedia's\n"See also" link graph). We evaluate MSA's performance on benchmark datasets for\nmeasuring semantic relatedness of words and sentences. Empirical results show\ncompetitive performance of MSA compared to prior state-of-the-art methods.\nAdditionally, we introduce the first analytical study to examine statistical\nsignificance of results reported by different semantic relatedness methods. Our\nstudy shows that, the nuances of results across top performing methods could be\nstatistically insignificant. The study positions MSA as one of state-of-the-art\nmethods for measuring semantic relatedness, besides the inherent\ninterpretability and simplicity of the generated semantic representation.\n